Upload
pembroke
View
29
Download
5
Embed Size (px)
DESCRIPTION
Factors Affecting the Success of Non-Majors in Learning to Program. Susan Wiedenbeck Drexel University. Introduction. End-user programmers increasingly important 55 million end users Programming spreadsheet formulas, macros, style sheets, web applications, domestic technologies - PowerPoint PPT Presentation
Citation preview
ICER 2005
Factors Affecting the Success of Non-Majors in Learning to Program
Susan Wiedenbeck Drexel University
Introduction End-user programmers increasingly
important 55 million end users Programming spreadsheet formulas, macros,
style sheets, web applications, domestic technologies
Programming education is a strong facilitator to end-user programming
But dropout and failure rate in introductory programming courses of 15-30% (Guzdial & Soloway, 2002)
How do cognitive and social-cognitive factors affect success and failure in an introductory programming course?
Factors Suggested in the Literature
Previous computer programming experience (secondary school)
Mathematics or science background Learning styles Course outcome expectations Playfulness (studied in training
courses) Self-efficacy
Factors Studied
Previous experience and two important factors in cognitive and social cognitive theory:
Self-efficacy: the individual’s judgment of his or her ability to perform a task in a specific domain
Knowledge organization: a person’s internal organization of concepts
Goals of this research Study previous experience, self-
efficacy, and knowledge organization of novice non-majors
Explore the relationship between these factors
Investigate their combined influence on course performance
Background on Self-Efficacy
Albert Bandura’s theory 1977 Basic Principle: “people are likely
to engage in activities to the extent that they perceive themselves to be competent at those activities”
Self-beliefs a key element in human performance
Self Efficacy and Cognitive Strategies
Self-efficacy influences behavior in poblem solving: effort expended persistence coping strategies adopted performance outcomes
Four Sources of Self-Efficacy
Performance attainments Experiences of observing the
performance of others Verbal persuasion Physiological reactions people use
to judge their capableness and vulnerabilities (anxiety, fatigue)
Knowledge Organization in Programming
Effective knowledge organization supports reasoning, comprehension, and explaining the behavior of systems Skilled programmers chunk knowledge into
meaningful groups in memory, recognize patterns, uses context of chunks to plan
Less skilled programmers don’t chunk related knowledge, each piece is an independent unit, doesn’t support pattern recognition and planning
PreviousExperience
Pre-Self-efficacy
Knowledge Organizatio
n
Post-Self-efficacy
Performance
The Proposed Model
Methodology - Participants
120 non-majors (34%F, 66%M) Average Age: 20 Second and third year undergraduates
from a variety of majors 5 sections of C++ programming
course 4 different instructors coordinated to
cover the same material
Methodology - Materials Background questionnaire Self-efficacy scale (Ramalingam &
Wiedenbeck, 1998) Knowledge organization
Program memorization/recall task (Shneiderman, 1976; Di Persio, Isbister, Shneiderman, 1980)
Indicator of well-organized knowledge Predicts exam grades
Methodology - Procedure
Two parts during one 16-week semester Part 1: Weeks 1-8
Background questionnaire Pre-self-efficacy score Program memorization/recall task
Part 2: Weeks 9-16 Post-self-efficacy score Debugging score on final examination Final course grade
Results - 1
Previous experience data
Mean
Number of courses using computers 2.62
Number of programming courses 1.15
Number of computer programs developed
~10
Number of programming languages 1.43
Longest length of program (lines) ~20
Results - 2
Mean StdDev
Pre-SE 3.73 1.54
Post-SE 5.11 1.17
Self-efficacy data – mean question response on scale of 1-7
Analysis of Model
Path Analysis Way to analyze predictive
relationships of variables in a model Uses series of multiple regressions Also concerned with analyzing the fit
of the overall model with the data
Results of the Analysis
PreviousExperience
.48**
.31**
.38**
.26**
.38**
.21**
..13
Pre-Self-efficacy
MentalModel
Post-Self-efficacy
Performance
R2=.43
R2=.27
R2=.02
R2=.24
**p<.01
Revised Model of Factors
PreviousExperience
Pre-Self-efficacy
Knowledge Organization
Post-Self-efficacy
Performance
.48**
R2=.24
.31**
.38**
.21**
.36**
.35**
R2=.43
R2=.30
R2=.02
-.20*
*p<.05
**p<.01
Discussion - 1 Self-Efficacy increased substantially
over the course of the semester Frequent hands on programming tasks Malleability of self-efficacy
Previous experience influences pre- and post-self-efficacy (indirect relationship to performance)
Previous experience did not affect students’ knowledge organization
Discussion - 2 Strong knowledge organization
increases beliefs in self-efficacy Self-efficacy and knowledge
organization influenced course performance
But high pre-self-efficacy combined with low previous experience was predictive of lower course performance
Limitations
The time span of a 16 week semester Self-reporting of previous experience Use of course grade as the
performance outcome What happened to the students who
dropped out during the semester?
Pedagogical Interventions - 1
Build a student’s self-efficacy Frequent hands-on programming
activities – learner-appropriate exercises with performance successes
Observation of the performance of peers –video or live observation of student working out a programming problem
Pedagogical Interventions - 2
Social persuasion – collaborative exercises, groups of students with varying degrees of abilities
Monitor students physiological state – design the classroom to decrease anxiety, reduce competition
ICER 2005
Thank you